July, 1929
THE DIELECTRIC PO1,ARIZATION OF LIQUIDS.
V
2051
contraction of density globes is a linear function of the pressure difference to which the bulb is submitted. BRUSSELS,BELGIUM [CONTRIBUTION
FROM THE
DEPARTMENT O F CHEMISTRY
OF PRINCETON
THE DIELECTRIC POLARIZATION OF LIQUIDS. ATOMIC POLARIZATION
V.
UNIVERSITY ]
THE
BY C. P. SMYTH RECEIVED hfARCEi 6, 1929
PUBLISHED JULY 5, 1929
I n previous communications’ the total polarization, P , of a substance has been divided into three parts, PE, the electronic polarization, PA, the atomic polarization and P>v, the orientation polarization due to the permanent moments of the molecules. This series of papers is concerned mainly with the conclusions to be derived from the values of P M for liquids, but, as PM is commonly obtained from the total polarization, P, by subtracting P E and P A , i t is evident that no treatment of the problem can be complete without consideration of the two latter quantities. As P E is discussed in another paper,2 the present treatment will be confined to the significance of PA, especially in its relation to molecular structure. The polarization is commonly obtained from dielectric constants measured a t frequencies up to 3,000,000 cycles or wave lengths down to 100 meters. For much higher frequencies or shorter wave lengt.hs, the orientation of the dipoles in an applied field often becomes less and P,,* diminishes, being negligible for frequencies in the infra-red region. In the range of visible light the frequency of the alternating field is so great that, for the most part, the electrons only are affected by it, the polarization being PE alone. P E is thus determined as the molar refraction for visible light. I n the absence of anomalous dispersion, the small variation of P E with wave length may be calculated by a simple dispersion formula such as that of Cauchy or Sellmeier3and the value of P E obtained for the comparatively low frequencies, virtually infinite wave length, a t which the dielectric constant is measured. However, the presence of oscillators with frequencies in the infra-red region causes anomalous dispersion-and, in consequence, polarizations greater than the values obtained by extrapolation with the simple formulas. Actually these oscillators are displaced in the electric field, forming electric doublets and thus contributing to the polarization of the medium. Although some of these oscillators may be electrons, the majority are atoms, ions or radicals and, for this reason, their contribution (a) Smyth, Morgan and Boyce, THISJOURNAL, 50, 1536 (1928); (b) Smyth and Morgan, ibid., 50, 1547 (1928); (c) Smyth and Stoops, ibid., 50, 1883 (1928). * Smyth, Engel and Wilson, THISJOURNAL,51, 1736 (1929). See Errera, “Polarisation Diblectrique,” Les Presses Imiversitaires de France, Paris, 1928; also Smyth, Phil. Mag., 45, 849 (1923).
:‘O.j:!
Vol. 51
C . P . SMYTH
to the total polarization is termed the “atomic polarization” and designated by P P A is usually so small and difficult to determine with accuracy4 that any attempt to subdivide it into the contributions of atoms, radicals, ions, etc., seems rather futile in the present state of our knowledge. Since P = P E f PA f P u and P and PE can always be determined, P.,is obtained by difference when PM is known. P M can be calculated from the temperature variation of the polarization of a gas or of dilute solutions, from which the polarization a t infinite dilution can be obtained, From measurements of dielectric c ~ n s t a n t ’in ~ )the ~ solid state a t temperatures sufficiently low and frequencies sufficiently high to prevent all orientation of the dipoles in the alternating field, a value can be obtained for the polarization from which PIv is eliminated, so that P A can be obtained by subtracting P E . Errera3 has assembled a considerable number of values of PA from the literature, calculating several not previously known. These are regrouped in Table I together with new values calculated by the writer and others taken from the literature. The values of the electric moment of the molecule, p , are shown in order that it may be ascertained whether there is any connection between h and PA. For the gases, PE PA is determined from the temperature variation of the dielectric constant and PA then obtained by subtracting PE. A similar method is employed for CHC13, C4H9C1,C2HjBr,C4H9Br,CzHJ and C6H5C1,the substances being measured in solution. For the liquids without moment, P A is determined by subtracting P E from P I and for the majority of those with moment by subtracting P E for the liquid from P for the solid. A small error in these calculations is caused by the fact that the value of PE for the liquid may often be slightly different from those for the solid and the vapor, which are usually unknown. I n the case of liquids for which ~1 is given as 0, it is sometimes impossible to distinguish between a very small value of L./ and 0, which means that PM may have a small finite value, instead of 0. This would, of course, cause a small error in PA. As, in general, P A is determined as a small difference between relatively large quantities, all the errors of experiment and method are accumulated in its values. A number of values obtainable from the literature, in particular several, for vapors, are obviously so much in error that they are excluded from Table I, and the decimal places for many have been rounded off. The literature sources from which the values of p and PA have been obtained are indicated by the numbers following their values. Where no number is given after P A , the value has been calculated by the writer from the data used in obtaining 1.1
+
Cf.Ebert, Z. physik. Chem., 113, 1 (1924). Debye, “Handbuch der Radiologie (Marx),” Akademische Verlagsgesellschaft m. b. H., Leipzig, 1925, Vol. VI, p. 619. 6 Errera, BirZZ. sci. acad. roy. Belg., [ 5 ] 12, 327 (1926). 6
July, 1929
THE DIELECTRIC POLARIZATION OF LIQUIDS.
V
2053
and the refractive indices in Landolt-Bomstein or "International Critical Tables." TABLEI ELECTRIC MOXENTSAND ATOMKC POLARIZATIONS PA
1 . 2 (2) 0 . 2 (2) .7 (2) .1 (2) 0 (6) . 5 (2) 0 (5) - 1 (2) 0 (2) . 9 (vapor) . 9 (2) .8 (2) 0 (2) . 4 (2) 1 . 9 (2) 0.65 (10) .88 (11) .91 (11) .96 (11) .92 (11) .83 (11) .78 (11) .90 (11) .88 (11) .76 (11) 1.09 (11) (4)
HCI HBr HI A pi2
3"
02
co coz csz so2
CH4 CZH6 C2H4 CzHz n-CsH14 n-C&e 2,2-Dimethylpentane 2,4-Dimethylpentane 2-Methylhexane 3-Methylhexane 2,2,3-Trimethylbutane 3,3-Dimethylpentane 2,3-Dimethylpentane 3-Ethylpentane 2,2,4-Trimethylpentane CHIC1 CHzC12 CHC13
CHZ=CC12 cis-CHCl=CHCl trens-CHCI=CHCl cis-CHCl=CHBr trans-CHCl=CHBr
{
6 . 2 (vapor) (12) 5 . 8 (liquid) (10) 6 . 0 (solid) (13) 3 . 0 (solid) (2) 3 . 0 (2) 5.7 (2) 10.6 (2) (5) 5.4 (11) (10) (25) (vapor) (28) (vapor) 12 (liquid) 2 . 8 (2) 3 . 4 (2) 3 . 2 (2) 3 . 6 (2) 3 . 5 (2)
2054
Vol. 51
C. P. SMYTH
TABLE I (Concluded) P x IOU cis-CHBr=CHBr trans-CHBr=CHBr cis-CHI=CHI trans-CHI=CHI Cyclohexane Benzene P-Ce&(CH& CeH5NO2 CaHsCl o-CeH4Clz W-CeH4C1? p-CsH4Clz o-C&f4BrZ m-C6H4Br2 o-CeH412
HzO
1.22 (2) 0 (2) 0.75 (2) 0 (2) 0 (2) 0 (17) 0 (18, 19) 3 . 9 (19) 1.52 (IO) 2 . 2 5 (20) 1 . 4 8 (20) 0 (2, 20) 1.67 (2) 1.22 (2) 1 . 6 3 (2)
PA
5.3 (2) 4.8 (2) 8.0 (2) 5 . 6 (2) 0.74 (2) 1 . 5 (2) 2 . 1 (2) 8 (2) 3 . 3 (IO) 5 . 8 (2) 4 . 4 (2) 3 (2) 3 . 8 (2) 2 . 2 (2)
i
5 . 3 (vapor) (2) 21’3 . 0 (solid) (2) (2) HzS 0.93 (6)
